SP22: Machine Learning for Signal Processing - Assignment 6

Name : Akhila Sakiramolla (asakiram@iu.edu)

UID : 2000886005

Importing required libraries

P1: Stereo Matching (revisited)

From the above histogram, I observe that there are 4 clusters.

Expectation Maximiztion algorithm

The expectation–maximization (EM) algorithm is an iterative approach for determining (local) maximum likelihood or maximum a posteriori (MAP) parameter estimates in statistical models that rely on unobserved latent variables. The EM iteration alternates between an expectation (E) step that constructs a function for the expected log-likelihood evaluated using the current estimate for the parameters and a maximization (M) phase that computes parameters that maximize the expected log-likelihood discovered on the E step. In the next E phase, these parameter estimates are utilized to determine the distribution of the latent variables.

(Reusing code from previous assignment)

E - step : calculate posterior probabilities

image.png

M - step : to update mean, variance and prior parameteres

image.png

ICM implementation

P2: Probabilistic Latent Semantic Indexing (PLSI) for Speech Denoising

PLSI Implementation

P3: PLSI for Analyzing Twitter Stream

Perceptron implementation